One of the biggest challenges in ICT-mediated learning models, such as e-learning and b-learning, is the process of skills assessment. Specifically during the elaboration of random questionnaires based on question banks in Learning Management Systems (LMS), where difficulties such as misspellings or conceptual errors arise. This can make questions difficult for students to understand, and therefore, the possibility of answering wrong by accident. In addition, sometimes questions are not organized by topics or levels of difficulty, which can lead to exams on topics not yet seen in class, or questionnaires with the wrong difficulty level among students in the same course. To reduce this problem, we developed a peer review tool for evaluating and adjusting questions and then exporting them to an LMS as a question bank, organized by topic and difficulty level. This work shows the structure of such a tool, and first validation results for impact and quality using double-blind testing in questionnaire design for a course taught at a higher education institution in Colombia.